Anthropic Releases Claude Fable 5.1: 52.6% Terminal-Bench-Science and 75% Cache Read Price Cut

On September 1, 2026, Anthropic officially released Claude Fable 5.1 and its safety-restricted counterpart Claude Mythos 5.1, updating its frontier intelligence tier three months after the initial Fable 5 rollout in June.

Claude Fable 5.1 ships with a default 1-million-token context window and up to 128,000 output tokens per request, with adaptive thinking permanently enabled. The model recorded a major performance breakthrough on scientific execution, scoring 52.6% on Terminal-Bench-Science 0.1 in Anthropic's published evaluations—more than doubling the 24.7% achieved by Fable 5 and outpacing Claude Opus 5's 29.0%. On general software engineering, the model achieved 55.8% on Terminal-Bench 4.0, taking the top spot on the Artificial Analysis Intelligence Index across all measured reasoning tiers.

Commercially, Anthropic retained base token rates at $10.00 per million input tokens and $50.00 per million output tokens while slashing prompt cache read pricing by 75%, dropping from $1.00 to $0.25 per million tokens. Anthropic estimates this lowers the effective run cost of typical agentic loops by 25% to 45%, prompting AI software engineer platform Devin to migrate its production Opus 5 workloads to Fable 5.1 on launch day.

The model is available immediately across Claude.ai for paid tiers (Pro, Max, Team, Enterprise), the Anthropic API, Amazon Bedrock, and Google Cloud Vertex AI under the model identifier claude-fable-5-1.